Paul Smith is EMEAI laboratory compliance productivity specialist at Agilent Technologies UK Ltd, 610 Wharefedale Road, IQ Winnersh, Wokingham, Berkshire RG41 5TP, United Kingdom, paul_smith@agilent.com

Paul Smith

Most companies have experienced being audited and, where necessary, “defending” the work carried out in their analytical laboratories during audits. Historically, laboratories have tended to provide information about the validation of their methods and procedures, the qualification and suitability of their analytical equipment, and information about training of their laboratory staff as justification for the validity of the analytical results.

Nonetheless, the focus on data integrity by FDA, the United Kingdom Medicines and Healthcare Products Regulatory Agency (MHRA), and other regulatory bodies during audits may mean that historical approaches to laboratory audit preparation and audit “defense” is simply not enough. In a data integrity-focussed audit, the emphasis has moved away from providing information based on technical justification and scientific rationale towards providing evidence that the analytical results are not fraudulent. This is almost a “guilty until proven innocent” approach and can be very different to historical audits. For any laboratories that are not prepared for this change, the audit will at best be “uncomfortable” and at worst may present a potential high risk to the organization.

FDA warning letters (1-3) and the recent announcement of MHRA (4) have highlighted the increasing focus on data integrity in the laboratory. Data integrity is a subject that many laboratories currently have significant concerns about (5). To add to those concerns, even the term “data integrity” can have widely differing meanings or interpretation, and there are currently few definitive reference sources available on the subject.


The acronym ALCOA has been widely associated with data integrity by FDA and was first used by Stan Woollen when he worked for the agency to help him remember compliance terms relevant to data quality (6). The good automated manufacturing practice (GAMP) guide “A Risk-Based Approach to GxP Complaint Laboratory Computerized Systems” (7) includes an appendix (Appendix 3) on data integrity. The terms used in the appendix are sometimes referred to as “ALCOA +” because they incorporate additional terms based on the European Medicines Agency’s concept paper on electronic data in clinical trials (8). The terms associated with ALCOA + are described as 

ccurate, complete, consistent, enduring, and available (see 

Who performed an action and when? If a record is   changed, who did it and why? Link to the source data.

Data must be recorded permanently in a durable medium   and be readable.

The data should be recorded at the time the work is   performed and date / time stamps should follow in order.

Is the information the original record or a certified true   copy?

No errors or editing performed without documented   amendments.

All data including repeat or reanalysis performed on   the sample.

Consistent application of data time stamps in the   expected sequence.

Recorded on controlled worksheets, laboratory notebooks   or electronic media.

Available / accessible for review / audit for the life   time of the record.

Comparisons are often made between secure electronic data and data that are available in paper format. The comparison results in similar conclusions that electronic data are more secure, more difficult to manipulate or change, and any changes are easier to detect (assuming that the software is technically compliant to 

21 Code of Federal Regulations (CFR) Part 11

 (9) and technical controls are appropriately implemented). On the other hand, changes to paper data, such as a printed chromatogram, are simpler to make, but much harder to detect. FDA has previously advised that defining paper records as “raw data” (the so-called typewriter rule) does not satisfy the predicate rules, that the industry has misinterpreted the 2003 21 

 Scope and Applications Guidance (10) and that “the printed paper copy of the chromatogram would not be considered a true copy” (11). Although this comment was made about chromatographic data, the principles have much wider implications.

The available evidence and the regulatory data integrity focus implies that paper is “bad” and that laboratories should implement fully electronic systems that move away from paper or even hybrid systems as soon as possible. However, with many laboratories concerned about data integrity and unsure what to do, this implementation strategy could be a potentially dangerous over-simplification. It could result in an over-concentration of resource “only” on ensuring security of electronic systems when there may be other significant risks.

FDA observations around computer security, sharing passwords, and/or not activating audit trails are not new, and certainly, if a laboratory finds itself in this situation, it should act immediately. For most laboratories, however, implementing a fully electronic approach may represent a significant change from what they actually do now. To put into perspective the scale of the change that might be required, many laboratories currently still define printed copies as their raw data despite FDA’s clarification in its guidance on records and reports (11). Additionally, many laboratories perform chromatography calculations manually or using software such as Microsoft Excel, rather than the chromatography data system (CDS). Manual calculations present one of the highest data integrity risks, even where traceability to the “source” data used is unambiguous. In extreme cases, FDA have identified instances where sample weighing practices could mean laboratories have calculated and entered the sample weight after the injections so that the result would pass specifications (12). Therefore, laboratories need to take a strategic approach to implementing data-integrity improvements.


  Where should the laboratory start preparing for data integrity? Some of the common issues that repeatedly come up in FDA warning letters are: 

 Where analysts share passwords, it is not possible to identify who creates or changes records, thus the A in ALCOA is not clear.

 The system configuration for the software does not adequately define or segregate user levels and users have access to inappropriate software privileges such as modification of methods and integration.

 Laboratories have failed to implement adequate controls over data, and unauthorized access to modify, delete, or not save electronic files is not prevented; the file, therefore, may not be original, accurate, or complete.

 Integration parameters are not controlled and there is no procedure to define integration. Regulators are concerned over re-integration of chromatograms.

 The record is not complete in this case. The definition of complete data is open to interpretation--see references 13 and 14 for a detailed analysis of FDA 483 observations on complete data (

 In this case, the laboratory has turned off the audit-trail functionality within the system. It is, therefore, not clear who has modified a file or why.
  


  To balance the focus on electronic data that data integrity tends to drive, a useful approach is to map the workflow within the laboratory, to identify and list all of the steps performed for each analytical technique (from sample receipt to approval of results) and each laboratory operation. For each step, the mapping should identify: 

The extent to which the process is manual or automated

The possible risks associated with the step (e.g., how could fraud be prevented or detected).

In some instances, this type of mapping may have already been performed (e.g., Lean Sigma, Six Sigma). Even if this is the case, it should be reviewed. One of the purposes of data-integrity auditing is to actively look for evidence of fraud, and previous mapping exercises may not support the change in focus.


  Using infrared identification as an example of an analytical technique, the steps, decisions, and actions performed when an analyst completes material identification by infrared spectroscopy have to be identified. For infrared, the high-level process steps associated with performing material identification are shown in 

Nevertheless, when these steps are examined in detail, it soon becomes apparent that many of the steps have data integrity and/or scientific compliance risks associated with them. 

 shows an expanded representation of the flow chart shown in 

. Fundamentally, the way the sample is prepared for examination by infrared has a significant influence on the quality of the spectrum obtained. 

Historically, laboratories may have “rejected” a spectrum before comparison with the reference based on sample preparation (provided this was documented in the analysts infrared training). However, in the data-integrity world, this approach would now be interpreted as the spectroscopic equivalent to a laboratory performing “test injections” on an HPLC system (1-2, 12) and will, therefore, be difficult to defend. Hence, although warning letters have been issued because the laboratory was renaming infrared spectra, there are also warning letters associated with poor use of the technique (see reference 15 and warning letter review).

For the software used to control the instrument and record the spectra, the system administrator may be required to show the auditor the “live” system and answer detailed questions about how audit trails are implemented, provide information on the security features of the software, and explain the system configuration options. Consequently, the choice of system administrator is no longer just a technical decision—they must be confident under the pressure of a data-integrity audit. 

The laboratory must be able to defend the conditions it uses to record spectra relative to pharmacopoeia requirements and explain how any differences found between spectra are managed.  Subjective tests need careful monitoring. If this is an electronic comparison, the software algorithm performing the comparison must be validated, while a manual comparison of spectra must be included in a documented training process.

Finally, are the electronic files secure or is the laboratory defining the print out as raw data? How can the audit trail be applied to detect changes to the spectra, and are the electronic files being examined when quality decisions are being made? 

Even for something as apparently simple as identification by infrared spectroscopy, when the process is mapped, a number of additional potential compliance risks can be identified, which might not be the case if the focus is only on the management of electronic data. Process mapping can, therefore, help identify “scientific validity” concerns from a data-integrity perspective, as well as highlight areas of high data-integrity risks.

, July 2013, accessed Apr. 9, 2014.
  2.	FDA, 

, Mar. 2014, accessed Apr. 9, 2014.
  3.	FDA, 

, Sept. 2013, accessed Apr. 9, 2014.
  4.	MHRA, “

MHRA expectation regarding self-inspection and data integrity

,”  accessed Apr. 9, 2014. 
  5.	Agilent compliance seminar, 

 (Milan, Italy, Apr. 2014).
  6.	S.W. Woollen, 

 in Newsletter of the Southern Regional Chapter Society of Quality Assurance, Summer 2010, accessed Apr. 9, 2014.
  7.	ISPE GAMP Good Practice Guide, “A Risk-Based Approach to GxP Compliant Laboratory Computerized Systems,” Second Edition, ISBN 978-1-936379-48-4, ISPE Publication, 2012.
  8.	EMA, GCP Inspectors Working Group publication,

 Reflection paper on expectations for electronic source data and data transcribed to electronic data collection tools in clinical trials

, Title 21, Food and Drugs (Government Printing Office, Washington DC,) 

Guidance for Industry, Part 11; Electronic Records, Electronic Signatures, Scope and Application

, (Rockville, MD, Aug. 2003). 
  11.		FDA, 

Guidance and Answers on Current Good Manufacturing Practices, Level 2 Guidance--Records and Reports, Question 3

, July 2013, accessed Apr. 9, 2014.
  13.		R. D. McDowall, LC-GC 

 (6) 338-343 (2013).
  14.		R. D. McDowall, LC-GC 

 (7) 389-392 (2013). 
  15.	P.A. Smith and J. Sellers, 


Paul Smith is EMEAI laboratory compliance productivity specialist at Agilent Technologies UK Ltd, 610 Wharefedale Road, IQ Winnersh,  Wokingham, Berkshire RG41 5TP, United Kingdom, 

Articles

Data integrity in the analytical laboratory is an area of increasing focus for regulators such as FDA.

Data Integrity in the Analytical Laboratory

Counterfeiting in the pharmaceutical industry is predicted to reach a value of approximately €52 billion in 2011 (1) and this threat has influenced the development of a range of measures to secure the supply chain and protect patients. One technology that can be useful for identifying counterfeit medicines is infrared spectroscopy, which provides a rapid fingerprint of information that allows a suspect drug to be compared with an authentic one. Indeed, the ease of sample preparation and rich content of an infrared spectrum means that infrared spectroscopy is "the first port of call in identifying counterfeit medicines" (2, 3). 

Interpreting the differences between infrared spectra requires a good understanding of the technique. For "routine" verification, where infrared is used to confirm the identity of a raw material or material produced from a validated manufacturing process, warning letters issued by the FDA have highlighted fundamental failures in meaningful use of infrared due to poor understanding of the technique (4, 5). 

Material identification is a fundamental strength of Fourier Transform Infrared (FT-IR) analysis and the technique can be used for many applications. Because quantitative analysis can be performed by infrared, measurement of the amount of material present is possible (6). To identify gaseous species in counterfeit drugs, hyphenated techniques such as Thermogravimetric (TG)-IR may be used. For drug distribution in a tablet, however, which is the property of a manufacturing process, infrared can be used. The United States Pharmacopoeia (USP) general chapter on "Spectrophotometric Identification Tests" <197> states: "The IR absorption spectrum of a substance, compared with that obtained concomitantly for the corresponding USP reference standard, provides perhaps the most conclusive evidence of the identity of the substance that can be realized from any single test" (7). The strength of this statement has given rise to much interest in FT-IR identification within the pharmaceutical industry. 

Although instrumentation and software for infrared spectroscopy have continued to evolve, the loss of specialist spectroscopy knowledge in companies is a concern. Coupled with confusion over what is required by different pharmacopoeias, this lack of understanding has led to regulatory compliance issues. According to FDA warning letters: "This practice is unacceptable and raises serious concerns regarding the integrity and reliability of the laboratory analysis conducted by your firm." (8) 

With sufficient experience and reference texts such as Bellamy, and Colthup et al. (9–11), an infrared spectrum can be interpreted to identify which functional groups are present in the compound under examination. In order to identify a material, however, its infrared spectrum must be compared to one previously recorded.  

Compliance with the respective national pharmacopoeia is a requirement when supplying pharmaceuticals in a country. There are at least 42 national pharmacopoeias (12), all of which assume that identification is performed by Chemical Reference Material or Certified Reference Spectrum. Some pharmacopoeias contain infrared reference spectra that can be used for "pharmacopoeial-compliant" identification. The European Pharmacopoeia (EP) supports the use of reference spectra, but does not include them, while the USP does not support identification by reference spectrum, making the clear statement in General Chapter <197> that the corresponding USP reference standard must be used and recorded "concomitantly" (7) Warning letters issued by the FDA include citations for not using USP reference materials, with one recent letter highlighting that the use of a reference spectrum from the Chinese Pharmacopoeia is unacceptable (13,14). Each pharmacopoeia requires the use of chemical reference materials from that particular pharmacopoeia.  

When a reference spectrum is used or generated within a laboratory, there are two essential requirements that must be satisfied for reduced compliance risk when using infrared. Firstly, the spectrum of the reference material should be of the appropriate quality—this should be verified and documented. The reference spectrum will be used to compare batches of material so it is essential that it is a good quality spectrum. The second requirement is that where an external reference spectrum is used for identification, it should be justified and documented by a supervisor or laboratory manager so that both the spectrum source and the appropriate use of the spectrum are approved by a qualified person. The first point is significant because traditional techniques of sample preparation for solids, such as the preparation of alkali halide disks (USP <197K>) or mineral oil mulls (USP<197M>) (7), are variable, and the quality of the infrared spectrum obtained will depend on the skill and experience of the analyst.  

One technology that has had a huge impact on sample preparation for infrared is high-pressure attenuated total reflectance (ATR) technology, which enables good-quality infrared spectra to be recorded on a wide range of materials in a short space of time. Additionally, for solid powders, the variations in spectra due to grinding or the application of high pressure are removed because quality spectra can be obtained directly by pressing the powder into contact with the ATR crystal. The increased productivity and consistency of this approach—the same technique can be applied to a range of material types—has transformed sample preparation. These advantages are highly applicable when checking for counterfeit drugs, as larger numbers of samples can be screened effectively. The USP actively supports the use of the newer ATR technology: USP <851> states: "A particularly common technique used for IR reflectance measurements is termed attenuated total reflectance (ATR)."(15) USP <197> adds: "The ATR <197A> and the <197E> techniques can be used as alternative methods for <197K>, <197M>, <197F>, and <197S> where testing is performed qualitatively and the Reference Spectra are similarly obtained." (7) 

However, this is not the case for the other pharmacopoeias. Of those reviewed in this article, the USP is the only one that documents the equivalency of ATR. 

Before FT-IR instruments existed, it was considered good science to record the infrared spectrum of a material in two stages: firstly, record and plot the sample spectrum and scan the chart back; secondly, overlay some of the peaks from a polystyrene sample. The partially plotted polystyrene peaks over the sample spectrum made it easy to see the accuracy of the wavenumber positions. For the Indian Pharmacopoeia (IP), this was a requirement as recently as the 2007 edition, but, as with the other pharmacopoeias reviewed in this article, this requirement has now been dropped.  

Although there is no simple ideal calibration standard, polystyrene is cited in pharmacopoeias, has been widely used and has certified reference standards readily available (16). An FT-IR spectrum of polystyrene is shown in 

compares the instrument requirements specified in a range of pharmacopoeias.  

Figure 1: Example a Fourier transform infrared spectrum of polystyrene (recorded at 4 cm-1 resolution). 			

The International Pharmacopoeia (Int. Ph) proposes the use of manufacturers' recommendations for instrument performance. The USP does not currently specify instrument performance criteria, but several FDA warning letters have been issued to companies for failure to calibrate FT-IR instruments (citing 21CFR211.160 (b)) (16–20). This is a commercially-sensitive issue for instrument manufacturers who supply to the US because customers often question whether an instrument is "USP compliant". Only the Japanese Pharmacopoeia (JP) includes a requirement to verify the reproducibility of the instrument. Where specified, there is good agreement between the pharmacopoeias for wavelength performance. 

Table I: Pharmacopoeia instrument performance requirements. 			

The polystyrene peaks referenced in wavelength calibration are shown in 

, along with their acceptance criteria. The JP, EP and IP currently include the weak, broad peak at approximately 1943 cm

  as a requirement in wavelength testing. This peak is associated with the overtones and combination bands of the aromatic ring (6), but was originally not a peak certified in the National Institute of Standards and Technology (NIST) 1921b standard. NIST now include this calibration peak in certified standards, but because of its weak, broad nature and greater uncertainty of measurement, its suitability as a calibration peak is questionable. NIST limit the use of this peak to certain peak position measurement algorithms. 

Some of the polystyrene peaks are sensitive to instrument resolution, which make them a good choice for resolution verification. However, a dispersive infrared will give over 12% transmission for resolution limit 2 in 

. For an FT instrument, this value is closer to 10% T at 4 cm

  resolution. The only way to guarantee a passing limit of 12% T for an FT-IR instrument is to scan at higher resolution. Historically, many pharmaceutical companies gathered evidence to support a 10% T value for limit 2 through a round-robin exercise of sending a polystyrene sample to many sites to record a spectrum, confirm the limit and justify the use of 10%T for this limit. In principle, recording all spectra at a higher resolution is simple and should not be a problem for FT-IR instruments. However, many commercial and historical spectra, including some of those in the pharmacopoeias, have been recorded at a resolution of 4 cm

 —scanning at a different resolution would cause compatibility problems. A scientific solution to this problem would be to change the values in the pharmacopoeias to what can be achieved at 4 cm

  in an FT-IR. Resolution limit 3 is only specified in the Chinese Pharmacopoeia (CP) and is satisfied at a resolution of 4 cm

 . In these instances, the International Conference on Harmonisation (ICH) guidelines of infrared requirements between Pharmacopoeias for infrared identification would be useful because they would clarify what was required, and help to ensure both a consistent approach and simplify understanding across the pharmacopoeias. 

compares the infrared identification requirements for six pharmacopoeias. A key point to note about all the pharmacopoeias is that where a sample spectrum contains differences relative to the reference spectrum being used, it is generally assumed that these may be attributed to the presence of another crystalline form, with typical wording (USP): "If a difference appears in the IR spectra of the analyte and the standard, dissolve equal portions of the test specimen and the reference standard in equal volumes of a suitable solvent, evaporate the solution to dryness in similar containers under identical conditions, and repeat the test on the residues" (15). 

Table II: Pharmacopoeia infrared identification requirements. 			

This fact in itself, or the resulting action that should be taken where spectra show a difference, is not universally known. One of the uncertainties is: how significant must the difference be before it is acted upon? There is no simple answer to this question; however, sample preparation, such as grinding, can affect the appearance of the infrared spectrum of a solid. This impact should be documented because it can help defend an auditor with a strong interest in infrared (18, 19). Additionally, examining infrared spectra in detail helps analysts to gain an understanding of what information a spectrum contains, which could help prevent instances where infrared is used incorrectly, or where the spectra between several chemically-related materials are identical (20–22). The use of dedicated identification software, specifically developed for the pharma industry, can also help to reduce the risk of such gross errors. 

The USP provides a mechanism for using ATR as an alternative to other forms of sample preparation. For other pharmacopoeias, the justification for using ATR as an alternative sample preparation technique must be provided by the company that wants to use this method (discussed later). Does this mean that ATR can be adopted within the pharmacopoeial requirements? In part, it depends on what level of detail a company has registered. The mechanism for making changes to registered processes depends on the level of the difference being made. There are generally three types of regulatory change approval: 					

major change—prior approval required (e.g., pre-approval inspection)

moderate change—period of notification required

minor change—make the change and then tell them.

A change of sample preparation would be classed as a minor change. In Europe, this would be a Type 1A change under the centralised new variation scheme (e.g., a minor change). In the US, it could be included in an annual update if identification by infrared had also specified sample preparation in the registration. 

However, what about using ATR instead of what is in the pharmacopoeias, such as the EP? At this point, it is worth noting that pharmacopoeia monographs are standards and not release tests (23). Where there is an appropriate scientific justification, the use of alternatives to the pharmacopoeias has been accepted. The three steps that would need to be put in place to develop ATR as an alternative sample preparation are: 					

Justification—to overcome problems of variation in sample preparation.

Documentation of equivalence between the tests—for identification.

For fundamental justification, the use of ATR for sample preparation overcomes the problems of recording good quality spectra by KBr disk or Nujol mull, and the fundamental inconsistency of variation of sample preparation (hand grinding the sample with a mortar and pestle in particular) between analysts. For the documentation of equivalence, it is necessary to use the method stated in the pharmacopoeia and then ATR, using the same reference material and sample. This would document the equivalence of infrared identification using ATR.   

FT-IR within the pharma industry is being increasingly performed by analysts who are not specialists and who tend to use the technique only occasionally (24). This subject was raised by UK spectroscopist Geoff Dent over 15 years ago when he wrote an open letter to Spectrochemica Acta on the subject of loss of infrared expertise (25). This trend has continued and has pushed instrument manufacturers to produce more powerful and user-friendly software. Software now includes verification that the ATR accessory is not contaminated (from the previous sample for example) and automated removal of atmospheric water vapour peaks. The infrared software can also be configured to perform material identification using spectral comparison. 

Application specialists at instrument companies have increasingly taken on the critical knowledge and technical role of the former in-house infrared spectroscopist. There is now a need for intelligent instruments that can perform self-diagnostic tests, receive a remote diagnostic connection for instrument fault determination, or even provide user-guided self installation to satisfy instrument qualification requirements. Software architecture can now include simple updates if limits in pharmacopeia requirements change. Additionally, the user-friendly interfaces with guided help and video tutorials mean that occasional users are much less likely to make errors. This, in turn, reduces the burden of compliance risk of a regulatory issue by the FDA during an inspection. Similarly, less experienced users can record infrared spectra for anti-counterfeiting applications while reducing the compliance burden on pharmaceutical companies with respect to regulated FT-IR testing for material identification. 

is validation programme manager (Europe) at PerkinElmer. 

is IR business manager, analytical sciences and laboratory services, PerkinElmer. 

 R. ColliaSuzuki, "The War Against Counterfeit Drugs", 

 P. Coombes, "Advances in technology in the fight against counterfeiting", BNET website (January/February 2011). 

 FDA, FDA Warning Letter KAN 2011-02 (8 November 2010). 

 FDA, FDA Warning Letter NYK 2011-09 (10 November 2010). 

 P. Smith, "Analysis of Crosslinked polystyrene based polymers using FT-IR/Raman Spectroscopy", MSc Thesis, (Birckbeck College, University of London, UK, 1989).  

 29 General Chapter <197>, "Spectrophotometric Identification Tests". 

 FDA, FDA Warning Letter WL-310-10-01 (28 January 2010). 

By providing an information "fingerprint", infrared red spectroscopy is a useful tool for identifying counterfeit pharmaceuticals. The authors identify best practices for ensuring compliance.

Paul Smith

Articles

Data Integrity in the Analytical Laboratory

FT-IR Identification: the Expertise Required To Ensure Compliance